Extragalactic Surveys

1
Extragalactic Surveys
Amy Barger
2

• Where Are We?
• Chandra/XMM have revolutionized distant AGN
  studies
• Now possible
• to map the history of a large fraction of the AGN
  population using hard X-ray surveys, and
• for the first time, to compare high-redshift
  low-redshift samples chosen in the same
  rest-frame hard energy (2-8 keV) band

3
Striking how modest the number of X-ray sources
is compared to the number of optical sources
4

• What Do We Need to Map the
• AGN History?
• Pyramid of surveys running from large area,
  shallow surveys to CDF-N level ultradeep surveys
  that is, as large a base as possible and as high
  a tip as possible
• One of the key issues can we justify needing
  even deeper fields, say up to 5 Ms? For example,
  to measure the faint end of the luminosity
  function in the z2-3 range?

5
AGN
Steffen et al. 2004
6
What Else Do We Need? Spectroscopic
identifications! Photometric redshifts are a
possible substitute---with the addition of NIR
and MIR data, we are able to make reasonable
photometric redshift estimates and use them to
construct luminosity functions
However, they do not tell about the spectral type
of the galaxy producing the X-ray light, and . . .
7
. . . there is danger in the scatter if a
phot-z scatters a source into a region where
there are not many objects (say, a catastrophic
error scattering a low-z source to z3), then one
can get the LF there very wrong GOAL have the
spectroscopic identifications be as complete as
possible (wide-field NIR spectrographs should
help for z1.6-2.6)
Redspec-zs Purplephot-zs Diamondsunid
Negative values mean nuclear dominated
8
Additional Needs Really important to have
complete wavelength coverage to understand the
spectral energy distributions of the AGN Also
want to know if there are other AGN sufficiently
thick that they are not being seen in current
hard X-ray surveys
9
Above f(2-8 keV)10-14 ergs cm-2 s-1, 80-90 of
hard X-ray sources have redshifts, while below
this flux, 60
ASCA
CDF-N
CLASXS
CDF-S
Barger et al. 2005
10
What Do We Mostly Agree On?
11
The Good News There is an astonishing amount of
agreement---we understand the X-ray samples very
well! But there still are issues that need to be
resolved
12
Uncertainties in the X-ray background measurement
are still at the 10-20 level, so we cannot
accurately determine the resolved fraction of the
XRB. This is a tricky issue, particularly for
population synthesis modelers who need to decide
what additional component to add in.
Hickox Markevitch 2006
13

• Source Classification
• One of the most important issues is that of
  source classification, and what we mean by the
  various classes
• Most groups use the four optical spectral classes
  of Szokoly et al. 2004, which are crude by the
  standards of optical AGN specialists
• absorbers,
• star formers,
• high-excitation sources (HEX),
• broad-line AGN (BLAGN FWHMgt2000 km/s)

14
Relative Contributions to 2-8 keV Light by
Spectral Class
33 BLAGN 7 HEX 27 (XBONG gt1042 ergs/s) 3
(OBXF lt1042 ergs/s) 30 Unidentified
Can now look at the X-ray colors by optical
spectral class
15
BLAGN are nearly all soft and show essentially no
visible absorption in X-rays, consistent w/our
understanding of them as unobscured
2 x 1021 cm-2
Barger et al. 2005
16
All the other AGN are well-described by a
power-law spectrum with photoelectric absorption
spread over a wide range of NH
3 x 1022 cm-2
Open squares---absorbers and star formers Solid
squares---high-excitation signatures Triangles---u
nidentified sources
Barger et al. 2005
17

• X-ray Luminosity Functions
• When computing rest-frame hard (2-8 keV) X-ray
  luminosity functions, one of the interesting
  things is to make a comparison with
  optically-selected QSO samples
• To do that, one needs to use the optical
  spectroscopic classifications to determine the
  BLAGN luminosity function separately

18
As move to higher z, all the sources are
increasing in L while the LFs are maintaining the
same shapes if drift the x-axis, plots look very
much the same from z0 to z1.2
Barger et al. 2005
19
The Steffen Effect BLAGN dominate the number
densities at the higher X-ray luminosities This
therefore says that almost all luminous objects
are unobscured, which instantly says there must
be some luminosity dependence on the obscuration,
since we know there is a substantial fraction of
obscured sources at the lower luminosities
Z0,0.4,0.8 shells
BLAGN
20
The shape of the BLAGN relative to the shape of
the total stays pretty much the same with z,
since both are obeying PLE This says that the
BLAGN fraction---that is, the ratio of the
integrals---stays the same over interval
z0-1.2 However, the objects that are BLAGN are
much less luminous at low-z than at high-z
Z0,0.4,0.8 shells
21

• Open Question
• This is a rather bizarre situation!
• What is it that leaves certain properties, such
  as the relative shapes of the two luminosity
  functions, so invariant, while changing the
  luminosities so much?
• In other words, why should a lower luminosity
  source be a BLAGN at lower redshifts, but a
  similar luminosity source not be a BLAGN at z1?

22
Higher Redshift Intervals Incompleteness larger
here, but phot-zs indicate unids mostly lie in
z1.5-3 interval Shapes no longer
well-represented by the maximum likelihood fits
to the z0-1.2 HXLFs computed at z1 (blue
curves) Thus, PLE does not continue beyond
z1.2 There are fewer low-L sources than one
expects, and so the light density is more
dominated by higher L sources at these redshifts
23
Good agreement between the optical and X-ray
selected LFs! Our optical spectroscopic
classification of BLAGN is consistent with that
of groups doing direct optical selection
Richards et al. 2005
24

• Classification Issue
• Up to now, we have just considered the optical
  spectral properties of the X-ray sources, but it
  is very reasonable to try to go the other way and
  ask,
• Just by looking at the X-ray properties, is it
  possible to tell whether a source is what an
  optical AGN specialist would call a BLAGN?

25
No, this is not so clear-cut. Have we optically
misclassified some sources w/soft X-ray spectra?
3 x 1022 cm-2
Open squares---absorbers and star formers Solid
squares---high-excitation signatures Triangles---u
nidentified sources
Barger et al. 2005
26
HEX sources are quite easily distinguished from
BLAGN and from the low excitation sources
weak Hb
narrow CIV
Noise is dominated by the noisiest spectrum
Cowie Barger
27
The low excitation sources have strong Hb and do
not show signs of NeV or of broad underlying
Balmer lines. 68 of the sources show no emission
lines at all
strong Hb
Noise is dominated by the noisiest spectrum
Cowie Barger
28

• Are BLAGN Being Lost in Other Ways?
• Are selection effects (e.g., galaxy dilution or
  spectral selection effects, such as whether
  spectrum includes Ha) causing one to misclassify
  BLAGN at low X-ray luminosities?
• Moran et al. 2002 Silverman et al. 2005
  Heckman et al. 2005
• For example could the absence of BLAGN at low
  X-ray luminosities be explained if the nuclear
  UV/optical light were being swamped by the host
  galaxy light?

29
Absorption line (in some cases, dont see any
UV nuclei)
z
Starbursts
LINERs Seyfert 1s
Seyfert 2s
ACS GOODS
BLAGN
30

• Well-known that the nuclear UV magnitudes and the
  X-ray fluxes for BLAGN are strongly correlated
• If galaxy dilution hypothesis were correct, would
  expect the non-BLAGN to be similarly correlated
  when we isolate their nuclear UV/optical light

31
Instead, turns out that, in general, the nuclei
of the non-BLAGN are much weaker relative to
their X-ray light than the BLAGN
Negative values mean nuclear dominated
32

• Thus, absence of BLAGN at low X-ray luminosities
  is not a dilution effect
• In general, non-BLAGN really have weaker
  UV/optical nuclei relative to the X-rays
• Thus, we are left with the situation that there
  is not a one-to-one correspondence---we cannot
  select only optical BLAGN just by looking at the
  X-ray properties

33
G. Hasinger SlideOptically Identified Hard
Samples
type-1 optical BLAGN, or galaxy with LXgt42,
HRlt-0.2 type-2 optical NLAGN, or galaxy with
LXgt42, HRgt-0.2
34
But then things get really confused! Sources
with soft X-ray spectra are -mostly BLAGN at
high X-ray luminosities -HEX sources become a
significant fraction at intermediate Ls -we do
not see AGN signatures in the optical spectra at
low Ls
35

• Conclude if one is going to split X-ray
  luminosity functions by class, one should do it
  based on optical spectral class alone or by X-ray
  color alone, but one should not try to mix them,
  because we do not understand how to relate one to
  the other
• But, maybe it would make sense just to split
  X-ray luminosity functions based on the X-ray
  colors alone, since that may be the best measure
  of whether a source is obscured or not

36
An X-ray color cut with G1.2 basically
reproduces the BLAGN LF at high Ls but adds in
sources at faint Ls, flattening out the LF, but
it is a little sensitive to where you place the
cut---a higher gamma cut makes it look much more
like the BLAGN LF. The Steffen Effect still
holds, however.
37
2-8 keV comoving energy density production rate
drops rapidly from z1 to z0 peak is in
interval z0.8-1.2
Openspectroscopic sources Solidall, including
phot-zs no ids put at z3
Negative values mean nuclear dominated
At zlt1.2, only 1/3 is due to broad-line AGNs
38
Cumulative growth of AGNs from Chandra (red
curve) compared with the cumulative SFH
Although both form most of their mass at late
times (z1), the AGN growth shows a slightly
different history is running later than the
SFH if AGN feedback has a significant effect,
the relative histories can help diagnose that.
Would like deeper images to check whether there
is a fainter z5-6 X-ray population. Potential
hidden gottcha presence of Compton-thick AGN.
39

• What Might We Be Missing?
• Deep MIR and radio images are an obvious avenue
  for searching for highly-obscured AGNs, since
  extinction in the MIR radio is small
• People have tried to use combined MIR radio
  selections, but to obtain a reliable upper limit
  on the possible population of X-ray undetected,
  obscured AGNs, a clean selection is needed
• Here we use a pure microJansky radio survey
  selection in the HDF-N field
• (207 sources to 40 mJy in a 310 arcmin2 area)

40

• The well-known correlation between radio power
  FIR luminosity makes it possible to estimate the
  total FIR luminosity of a galaxy from its radio
  power (this correlation has been empirically
  determined for both star formers radio-quiet
  AGNs)
• Even with Spitzer, this is still the most robust
  way
• limited MIPS sensitivities at 70 and 160 mm

41
-the conversion from 24 mm to total FIR
luminosity depends strongly on the template
spectral energy distributions used to K-correct
the data
Barger et al. 2007 (astro-ph/0609374)
42
What fraction of the X-ray light is coming from
the different X-ray and radio populations?
43
It is sufficient for us to measure the X-ray
surface brightnesses from known X-ray sources in
the CDF-N see that they are roughly consistent
with the HEAO1/A2 XRB (Revnivtsev et al. 2005)
Z0,0.4,0.8 shells
Barger et al. 2007
44
X-ray sources that are not radio sources and the
total radio sample each contribute 50 of the
X-ray light at 4-8 keV, but most of the radio
sample contribution is from X-ray luminous radio
sources
Z0,0.4,0.8 shells
Total radio
X-ray sources that are not radio
Barger et al. 2007
45
Contributions from the remaining radio sources
are very small
Z0,0.4,0.8 shells
Radio sample without X-ray luminous counterparts
(2.3 at 4-8 keV)
Radio sample without any X-ray counterparts (1.2
at 4-8 keV)
Barger et al. 2007
46
Thus, the current radio source population cannot
account for the background light that has been
suggested may be missing at 4-8 keV Indeed, the
percentages are about a factor of 5 lower than
those predicted for 1024 cm-2 sources at these
energies in typical XRB synthesis models (e.g.,
Gilli, Comastri, Hasinger 2006)
47
Summary

• X-ray data are consistent with many of the
  non-BLAGN (the dominant population) having high
  column densities
• AGN evolve very rapidly to z1.2, consistent with
  pure luminosity evolution to that redshift
• z1 is where the bulk of the supermassive black
  hole population forms
• Simple unified model is not correct, whether one
  uses X-ray color or optical spectral
  classification---there are far fewer low X-ray
  luminosity, unobscured sources than obscured
• Contributions to the 4-8 keV light from the X-ray
  faint radio population is very small, and hence
  these sources are unlikely to contribute
  substantially to the XRB at even higher energies

48
The End